STUDY OF AIR POLLUTANTS
BY MICROWAVE SPECTROSCOPY
Contract No. CPA-22-69-144
Final Report
June, 1972
George F. Crable
Analytical Laboratories
The Dow Chemical Company
Midland, Michigan 48640
Environmental Protection Agency
Research Triangle Park
North Carolina 27711
Dr. James Hodgeson, Project Officer
-------
FINAL REPORT
STUDY OF AIR POLLUTANTS
BY MICROWAVE SPECTROSCOPY
Contract No. CPA-22-69-144
Environmental Protection Agency
Research Triangle Park
North Carolina 27711
Dr. James Hodgeson, Project Officer
-------
-11-
SUMMARY
Background information and theory, particularly as they apply to
analytical applications are presented. Three possible analytical
procedures are discussed in some detail. These are the optimum
power saturation method, line areas, and line intensities.
The optimum power method has the advantage that the problem of
varying line widths is completely eliminated at pressures above
some minimum value. It has two important disadvantages: (1) the
power level must be optimized for each line and sample pressure
encountered, and (2) the amount of power required is well beyond
the capabilities of most microwave sources for most molecules.
Provisions must be made to operate the crystal detector of the
spectrometer so that its output is proportional to the square
root of the applied power.
The line area is proportional to the partial pressure of an ab-
sorbing molecule and is also independent of the line width. It
is however somewhat tedious to measure. We have shown that the
product of the line height and the half power width has the same
properties as the line area. Measurement is relatively quick
and easy. A disadvantage of both area methods is that the scan-
ning rate of the spectrometer must be uniform and repeatable.
This was accomplished by locking the frequency of the source to
a harmonic of a "tuned" crystal oscillator.
Direct quantitation by means of the line intensity is a very
simple and adequate method for certain cases of interest. These
include any system in which the desired component is in low
concentrations and the composition of the bulk of the mixture
is relatively constant. Trace components in air is an ideal
system for this method.
-------
-111-
The problem of adsorption of gases on the walls of the absorption
cell is serious. The amount of such adsorption to be expected
was shown by several experiments in which gas pressure was mea-
sured as a function of time. Equipment was assembled and a
study conducted to determine whether the adsorption problem
could be eliminated by a continuous flow-through system. The
results indicated that strongly adsorbing compounds can be
handled easier through preconditioning of a standard batch cell.
-------
-IV-
TABLE OF CONTENTS
Page
I. INTRODUCTION 1
II. MICROWAVE SPECTROSCOPY BACKGROUND FOR 4
ANALYTICAL APPLICATIONS
III. DESCRIPTION OF THE MICROWAVE SPECTROMETERS 8
IV. SPECTRA, COMMENTS, AND SENSITIVITY LIMITS 11
FOR THE COMPOUNDS STUDIED
V. QUANTITATIVE ANALYSIS BY THE OPTIMUM POWER 27
SATURATION METHOD
A. THEORY OF THE OPTIMUM POWER SATURATION METHOD ... 27
B. EXPERIMENTAL RESULTS FOR THE OPTIMUM POWER .... 30
SATURATION METHOD
C. SUMMARY AND CONCLUSIONS ON THE OPTIMUM POWER ... 35
SATURATION METHOD
VI. QUANTITATIVE ANALYSIS BY LINE AREA MEASUREMENTS ... 36
AND LINE HEIGHT MEASUREMENTS
A. DISCUSSION OF QUANTITATIVE ANALYSIS BY AREA .... 36
MEASUREMENTS
B. QUANTITATIVE ANALYSIS FROM LINE HEIGHT MEASURE- . . 37
MENTS
C. EXPERIMENTAL RESULTS 38
VII. GAS BLENDS AND THE ADSORPTION PROBLEM 47
VIII. MEASUREMENTS WITH A FLOW-THROUGH CELL 53
A. DESCRIPTION OF THE EQUIPMENT AND PROCEDURE .... 53
B. EXPERIMENTAL RESULTS WITH THE FLOW-THROUGH .... 54
SYSTEM
C. CONCLUSIONS 61
IX. SUMMARY 64
X. CONCLUSIONS 66
-------
-v-
LIST OF TABLES
Page
I - OCS 13
II - S02 14
III - Formaldehyde 16
IV - N02 17
V - CH3SH 18
VI - Acetone 21
VII - Potential Interferences with Selected Lines .... 23
VIII - Detection Limits in Mole Percent 26
IX - Microwave Power Required for Optimum Power .... 35
Saturation
X - Argon in Nitrogen Blends 47
XI - Blends of S02 in Nitrogen 48
XII - S02 in Nitrogen 49
XIII - Height of S02 22482.5 MHz Line and Time to .... 55
Reach Maximum Amplitude
-------
-VI-
LIST OF ILLUSTRATIONS
Page
Figure 1 - OCS 24326 MHz Intensity at Optimum 31
Power Saturation vs. Pressure
Figure 2 - OCS 24325.9 MHz Intensity vs. Pressure .... 39
Line Area vs. Pressure
Figure 3 - S02 24039.6 MHz Line Height vs. Pressure ... 40
Height x Width vs. Pressure
Figure 4 - CH20 22965.6 MHz Line Height vs. Pressure . . 41
Height x Width vs. Pressure
Figure 5 - Acetone 21169.2 MHz Line Height vs 43
Pressure Height x Width vs. Pressure
Figure 6 - CH3SH Line Intensities vs. Pressure 44
Figure 7 - CH3SH Height x Width x Pressure 46
Figure 8 - Pressure Decay of CH3SH in Cell of 51
Spectrometer No. 1
Figure 9 - Pressure Decay of CH3SH in Cell of 52
Spectrometer No. 2
Figure 1O- S02 in Nitrogen 22482.5 MHz Total 56
Pressure 100 mTorr Power O.1 mw
Figure 11- CH3SH in Nitrogen 25291 MHz Total 57
Pressure 100 mTorr Power 0.1 mw
Figure 12- Acetone in Nitrogen 21170 MHz 59
Total Pressure 100 mTorr Power 0.1 mw
Figure 13- CH20 in Nitrogen 22965.7 MHz 60
Total Pressure 1OO mTorr
Figure 14- NH3 in Nitrogen 23870.1 MHz 62
Total Pressure 100 mTorr Power 5 mw
Figure 15- O.l% N02 in Nitrogen 26569.2 MHz 63
Line Height vs. Time Total Pressure
100 mTorr Power 5 mw
-------
-1-
I. INTRODUCTION
The purpose of the work supported by contract number CPA-22-69-144
was to evaluate the usefulness of microwave spectroscopy as an
analytical tool for air contamination studies. For this type
of work, an ideal analytical technique would (1) provide positive
identification with little interference, (2) be applicable to
a large range of compounds, (3) be suitable for good quantitative
results, and (4) have high sensitivity. In addition, a desirable
feature would be the ability to examine an air sample directly in
the atmosphere, and at some distance from the particular analytical
instrument.
Microwave spectroscopy obviously does not qualify 100% in all
of the criteria given above. Its ability to provide a positive
compound identification is certainly unexcelled by any known
technique. In fact, the fundamental theory of microwave spec-
troscopy suggests the possibility that no future technique will
be developed with a greater degree of compound specificity. In
many cases, the identification by frequency of a single observed
line in a sample is sufficient for compound identification.
Finding two lines with matching frequencies and the proper in-
tensity ratio for a compound is a good positive identification,
while matching three lines to the spectrum of a given compound
makes the identification an absolute certainty.
The number of compounds which have adequate microwave spectra
is limited. The primary limitation is the requirement that a
molecule must have a permanent dipole moment of reasonable value.
Many molecules with small dipole moments have observable micro-
wave spectra, but only with specialized techniques and very slow
scans. The sensitivity for such compounds is generally too low
for practical analytical applications. The dipole moment re-
quirement is a necessary but not sufficient condition for a
-------
-2-
molecule to have a useful microwave spectrum. Large molecules
with many possible degrees of internal rotation have microwave
spectra consisting of a very large number of weak lines. How-
ever, in spite of these limitations, a number of small molecules
of interest to air contamination do have good microwave spectra.
The quantitative application of microwave spectroscopy has not
been discussed extensively in the published literature. The
quantitative handling of microwave spectral data is somewhat
unique compared to other forms of molecular spectroscopy. For
example, it is relatively easy to have too much source power in
a microwave spectrometer and thereby produce line-broadening
through power saturation. Line widths, and thus line intensities,
are also quite sensitive to the types of molecular collisions
which occur. The line width and intensity observed for a com-
pound of fixed mole percentage in a gaseous mixture will vary
with the nature of the other compounds which make up the mixture.
Part of our work in this contract was to evaluate quantitative
analytical methods using line heights directly, line heights
under power saturation conditions, and line areas. A complete
discussion of the advantages and disadvantages of these methods
will be given later.
The theoretical sensitivity, or limits of detection, of a micro-
wave spectrometer ranges from about 2 ppm for NH3 to a more
typical figure of 0.1 to 1.0 mole % as an average for a fairly
representative collection of molecules. The figure of 2 ppm
of NH3 is based on a spectrometer capable of detecting an ab-
sorption line of intensity 10 9 cm x, and a line intensity of
1.5 x 1O~4 cm"1 for the NH3 24139 MHz line. A number of small
molecules of interest have detection limits in the tens to
hundreds of parts per million range.
-------
-3-
Microwave spectroscopy is not useful for the direct observation
of gaseous contaminants in the atmosphere. A microwave spec-
trometer normally operates at a pressure of 10 to 100 millitorr
to keep the lines narrow. Increasing pressure produces line
broadening, and at atmospheric pressure the lines have been
broadened to the point where they can no longer be observed
as discrete lines.
The main body of this report will be concerned with a discussion
and evaluation of the experiments carried out under this con-
tract. The evaluations will be done in terms of microwave
spectroscopy as a practical quantitative analytical technique.
-------
4
II. MICROWAVE SPECTROSCOPY BACKGROUND
FOR ANALYTICAL APPLICATIONS
Microwave spectroscopy is the study of molecular processes which
result in the absorption of electromagnetic energy in the fre-
quency range of approximately 1000 MHz to 300,000 MHz. The
principal molecular process which produces such absorption is
the excitation of rotational levels of the molecule. Additional
absorption lines result from the coupling of nuclear quadrupoles
with rotational levels, internal rotation of one part of a mole-
cule with respect to another part, rotational levels within
excited vibrational states of a molecule, molecular inversion,
very low energy vibrational states, and others. Fortunately,
analytical applications of microwave spectroscopy do not depend
upon a knowledge of the process or levels involved in producing
a particular absorption line.
The application of microwave spectroscopy to qualitative iden-
tification problems does require a cataloging of all absorption
lines observed in the spectrum of a compound. However, the
identification of a reasonably pure compound requires only a
knowledge of a few of the most intense lines in the spectrum
of the compound. A search for minor components first requires
that all small peaks associated with the already identified
major components be identified and removed from consideration.
Minor component identification then proceeds from the remaining
peaks.
The rotational energy levels of a molecule are determined by
the principal moments of inertia of that molecule. Thus, the
rotational levels and absorption frequencies are defined by
the masses of all atoms in a molecule and the geometric position
of these atoms. A change in mass of an atom, e.g., by isotopic
-------
-5-
substitution, will produce a major change in the frequencies
of all lines in the spectrum of that molecule. A change in
geometrical arrangement of atoms within a molecule, such as
cis- and trans-isomers, also results in a major change in all
frequencies in the microwave spectrum. The dependence of ab-
sorption frequencies on the molecular moments of inertia makes
it essentially impossible for two molecules to have identical
spectra. This condition reduces, but does not eliminate the
possibility that two compounds may have absorption lines of
the same frequency. The unique spectra plus inherently high
resolution of a microwave spectrometer give microwave spec-
troscopy the ability to make an unambiguous compound identifi-
cation.
A fundamental requirement for the observation of microwave
spectra is that the molecule must possess a permanent dipole
moment. Molecules such as methane, benzene, CC14, and N2 cannot
be observed. Many molecules which produce good infrared spectra
from an oscillating bond moment, e.g., H-C-C-H or N-C-C-N,
have no microwave spectra because of the lack of a permanent
molecular dipole moment. The intensities of the spectra of
geometric isomers may be quite different because of differences
in the values of the permanent dipole moments of the two iso-
mers. For example, the cis form of 1.2 difluoroethylene has
a relatively intense spectrum, while the spectrum of the trans
form is quite weak.
An understanding of the effects of gas pressure and microwave
power on line intensities and shape are essential to quanti-
tative applications of microwave spectroscopy. The very small
energies involved and the ability of a microwave spectrometer
to observe the effects of such small energy changes makes
microwave absorption lines very sensitive to these effects.
-------
-6-
Absorption involves a transition between two energy levels.
In the presence of microwave energy, molecules in the lower
level absorb energy and move to the upper level while, simul-
taneously, molecules in the upper level are stimulated to
move to the lower level, and in the process they lose energy
by radiation. A net absorption of microwave energy occurs
because the Boltzmann distribution results in a slightly
higher molecular population in the lower level.
Collisions among the molecules are constantly acting to restore
the original Boltzmann distribution, while the microwave energy
acts to equalize the populations of the two levels. The col-
lision rate is directly proportional to pressure (for a fixed
composition of the gas). At low pressures, the collision rate
is small and a relatively small amount of microwave power will
cause the populations of the two levels to approach equality.
This result is power saturation. With a further increase in
microwave power, the line height does not increase; the line
becomes broader. As gas pressure is increased, the collision
rate increases and a larger amount of microwave power is re-
quired to produce power saturation.
Molecular collisions and power saturation cause the observed
microwave line to broaden with a resultant loss in line height.
The collision rate increases with gas pressure and thus the
line width increases with pressure. At some pressure in the
range of 10 to 2O mtorr, an increase in sample pressure produces
essentially no increase in line intensityonly additional
line broadening.
From line width measurements, collision rates and effective
collision diameters can be determined. In turn, from collision
diameters information on the type of molecular interactions,
-------
-7-
i.e., dipole-dipole, dipole-quadrupole, etc., can be evaluated.
To the analyst, line width measurements are important because
they allow him to estimate the usefulness of a particular quan-
titative procedure.
The overall effect of temperature on microwave lines is a
reduction in line height with increasing temperature. If we
compare identical gas samples at the same pressure but different
temperatures, the higher temperature sample will have (1) a
smaller population difference between a given pair of energy
levels, (2) a higher collision rate, and thus a wider line, and
(3) a smaller total number of gas molecules. All three effects
act to decrease the observed line height. For maximum sensi-
tivity a microwave spectrometer should be operated at the lowest
possible temperature.
-------
-8-
III. DESCRIPTION OF THE MICROWAVE SPECTROMETERS
Two spectrometers were used to obtain the experimental results
of this report. Both spectrometers are equipped to operate in
K-band (approximately 17,500 MHz to 27,OOO MHz) and, except
for equipment details, are very similar. Spectrometer no. 1
is our original spectrometer, while spectrometer no. 2 was con-
structed later and was designed as a more versatile instrument
for analytical development work.
The gas cells are Stark cells of X-band wave guide with a metal
Stark septum centered in the guide and parallel to the broad
faces of the wave guide. The septum is insulated by strips of
Teflonฎ along the two narrow sides of the wave guide. Each cell
is 10 feet long. The cell for spectrometer no. 1 is brass,
while the cell for spectrometer no. 2 was constructed entirely
of stainless steel. In addition, the cell of spectrometer no. 2
was constructed with pumping ports sections at both ends of
the cell to permit samples to be pumped through the cell in a
continuous flow-through system. Cell no. 2 is designed to oper-
ate at temperatures up to 250ฐ.
The square wave Stark modulation voltage for both spectrometers
is obtained from Industrial Components Incorporated square wave
generators which operate at either 1OO kHz or 5 kHz and produce
a peak voltage of 2000 volts. The phase lock amplifiers are
the Electronics, Missiles and Communication, Inc., model RJB.
The 100 and 5 kHz preamplifiers were constructed at Dow. The
crystal detectors are selected type 1N26C (Sylvania) crystals
mounted in Microwave Associates K-band crystal holders.
Microwave power levels were measured by means of a Hewlett
Packard model 431B power meter. When microwave power leveling
-------
-9-
was required, a leveler which was designed and constructed at
Dow was used. This unit levels power by automatically adjusting
the input power to the absorption cell to maintain a constant
voltage across the crystal load resistor and thus, a constant
microwave power level at the crystal detector.
The sample introduction system of spectrometer no. 1 is equipped
for direct introduction of gases through quick-couple connectors.
Both gases and liquids may be introduced through a silicon rubber
septum by means of microsyringes. The sample introduction sys-
tem of spectrometer no. 2 was specifically designed for optimized
handling of analytical samples. It was modeled after mass spec-
trometer inlet systems constructed here at Dow. The entire sys-
tem can operate from room temperature to 250ฐC. Samples can
be introduced by microsyringes through a rubber septum, by direct
introduction through a quick-couple connector, or solids and
liquids can be placed in cups of Teflonฎ and loaded through a
vacuum lock system. Both systems are equipped with spherical
glass ballast volumes to reduce gas adsorption effects on pres-
sure.
Diaphragm capacitance type micromanometers are used to measure
sample pressures in both spectrometers. Spectrometer no. 1 is
equipped with an Atlas-Werke, A.G., model EW230 micromanometer
which measures pressures from 0 to 200 mTorr, while spectrom-
eter no. 2 is equipped with an MKS Instruments, Inc., model
78M-XRP capacitance manometer with a range of 0 to 10 Torr in
eight ranges which are selected by a switch.
The microwave source for spectrometer no. 1 is a modified Alfred
backward wave oscillator power supply and a Varian Associates
model 163 backward wave oscillator. Sweep controls were con-
structed at Dow. The source for spectrometer no. 2 is a Hewlett
-------
-10-
Packard model 8690A source control with reference oscillator,
synchronizer and counter for direct read-out of frequency. A
Hewlett Packard model 8696A backward wave oscillator is used.
Approximate frequencies were measured with Hewlett Packard
model K532A cavity-type wavemeters. More exact frequency
measurements were based on harmonics generated by crystal os-
cillator and frequency multiplier units manufactured by the
Micro-Now Instrument Co. 50 and 450 MHz signals from the fre-
quency multiplier were mixed and applied to a harmonic gener-
ating 1N26C crystal which was mounted on a combination directional
coupler and attenuator in the plumbing from the source to the
cell. A parallel connection to the crystal was made to the
input of a radio receiver. With the receiver tuned to 25 MHz,
marker output from the receiver occurs every 50 MHz at those
times when the microwave frequency of the source and a harmonic
generated at the crystal differ by 25 MHz. The receiver output
is applied to the "Zero Beat In" terminals of the phase lock
amplifiers. With proper polarization of the connections, sharp
frequency markers are produced every 50 MHz on the spectral
chart and in a direction opposite to that of the absorption
line. Frequencies of the markers end in -25 or -75 MHz, such
as 22225, 22275, 22325 and 22375. Frequencies accurate to ฑ
1MHz can be obtained by linear interpolation between two frequency
markers. This system is simple to use, requires no prior knowl-
edge of line frequencies to be expected, and the accuracy is
more than adequate for line identification in practical analyti-
cal applications. For more accurate frequency measurements,
more elaborate (and more time-consuming) techniques are avail-
able.
-------
-11-
IV. SPECTRA, COMMENTS, AND SENSITIVITY LIMITS
FOR THE COMPOUNDS STUDIED
The compounds selected for study in this program were S02, NH3,
CH20, N02, CH3SH, and Acetone. A considerable amount of data
were also obtained for OCS because of its high intensity and
ease of handling. These compounds cover a range of types and
problems of potential interest in air pollution studies. Car-
bonyl sulfide has a very large absorption coefficient and is
a linear molecule; thus it has a very strong and uncomplicated
microwave spectrum. It is also generally used as a calibration
standard.
Sulfur dioxide has a strong and relatively uncomplicated spectrum
and is of interest because it is found in vent stack emissions
from power plants. Methyl mercaptan has a reasonably strong
microwave spectrum. It is rapidly adsorbed on the walls of an
absorption cell and provides an interesting model for adsorption
problem studies. Formaldehyde was chosen as a representative
oxygenated compound typical of automotive exhaust emissions.
Acetone is an example of a compound with less intense microwave
lines but a very complicated spectrum with hundreds of ab-
sorption lines.
N02 was included as a compound of great interest in automobile
emissions and other pollution areas. The problem in working
with NO2 is its high chemical reactivity. At the start of this
project, we questioned whether meaningful measurements could be
made for N02. Certainly the problems of reactivity and ad-
sorption make a quantitative determination of N02 at low levels
very unreliable.
NH3 is of interest in any study of analytical applications of
microwave spectroscopy because of its very strong absorption,
-------
-12-
possibly the strongest microwave absorber of all molecules. In
principal, the sensitivity of a microwave spectrometer for NH3
should be the highest of all molecules. However, its strong
adsorption to the cell walls of a spectrometer causes the loss
of NH3 from samples of low concentration. A trace of NH3 in
a sample can easily disappear on the cell walls. A further
complication is the appearance of small amounts of NH3 from
compounds containing amine groups. Every amine type of compound,
including aniline, that we have examined has shown an easily
observed spectrum of NH3. In mixtures of amines and NH3, the
source of the NH3 would always be in doubt.
Tables I through VI give the spectral data that we have observed
for OCS, S02, CH20, N02, CH3SH, and a part of the spectrum of
acetone. Our catalog for acetone includes 205 measured line
frequencies. These are only a portion of all the lines observed
in the complete spectrum. No tabulation was given for NH3 since
a rather complete listing with intensity data is given in NBS
Circular 518, Molecular Microwave Spectra Tables. The observed
frequencies in the tables were determined by interpolating
between frequency markers separated by 5O MHz. Frequencies
reported in the literature, or measured accurately here at Dow,
are included in the second column to give a measure of the
accuracy of the survey measurements using markers. The Dow
data reported under literature frequencies is believed to be
accurate to at least ฑ0.02 MHz.
The intensity data of Tables I through VI are all corrected to
OCS as a standard, except for a difference in microwave power
level. The data of the first three tables were obtained with
a power level of 0.1 mw while Tables IV, V, and VI were obtained
with l.O mw.
-------
-13-
Table I
Observed
Freq.(MHz)
22820
23661
23731
23761
23785
24014
24021
24248
24254
24326
24356
24381
24401
ocs
Power = 0.
Literature
Freq. (MHz)
23660. 62( *
23731. 30( ^
23760. 67 ( )
23784. 95(4)
24253. 51 (1)
24325.92*
24355.5O(1)
24381. 07 ( )
1 mw
Intensity
(divisions)
6.5
2.5
182
7.0
7.3
3. 5
19. 5
39
58
455O
183
183
26
Intensity,'
0.1
0.05
4.0
0.2
0.2
0.1
0.4
0.9
1.3
100
4.0
4.0
0.6
^Frequencies measured at Dow to an accuracy of better than
ฑ0.02 MHz.
-------
-14-
Table II
SO,
Power = 0.1 mw
OCS 24326MHz line = 455O div.
Observed
Freq.(MHz)
17971
18569
18977
19096
19227
19230
19250
19306
19636
19681
20103
20261
20335
20384
20548
20701
20785
20835
21265
21762
21768
22065
22220
22482
22734
22904
22929
Literature
Freq.(MHz)
Intensity
(divisions)
Intensity,%
20335.43*
22220.32
22482.51*
22733.83*
22904.95*
22928.45*
17
4
2
2
2
6
2
4
221
17
3
6
721
6
34
18
3
1
15
2
2
77
51
822
61
64
47
1.4
0.4
0.2
0.2
0.2
0. 5
0.2
0.4
17.9
1.4
0.2
0.5
59
0. 5
2.8
1. 5
0.2
0.1
1.2
0.2
0.2
6.2
4.1
67
5.0
5.2
3.8
-------
-15-
Table II
Observed
Freq.(MHz)
23035
23207
23415
23733
24040
24083
24302
24319
24796
24916
25050
25171
25393
25884
26038
26412
S02 (Continued)
Literature
Freq. (MHz)
23034. 80 ( )
23414. 25( )
23733. 03 ( )
24039.65*
24083.46*
Intensity
(divisions)
221
10
620
45
853
550
4
174
19
17
430
45
1232
11
31
90
Intensity,%
18
0.8
50
3.7
69
45
0.3
14.1
1.5
1.4
35
3.7
100
0.9
2.5
7.3
*Frequency measured at Dow to better than ฑ0.02 MHz.
-------
-16-
Power = O.I mw
Table III
Formaldehyde
OCS 24326 MHz line = 4550
Observed
Freq.(MHz)
18841
19595
20426
20650
20735
20794
22002
22966
24068
24731
26331
26362
Literature
Freq.(MHz)
19595.23
22965. 57*
24068.35*
26358.82
Intensity
(divisions)
2
73
1
12
11
2
5
982
831
2
2
15
Intensity,%
0.2
7.4
0.1
1.2
1.1
0.2
0. 5
100
85
0.2
0.3
1.6
^Frequencies measured at Dow to an accuracy of better than
ฑ0.02 MHz.
-------
-17-
Table IV
NO,
Power = 1 mw
DCS 24326 line = 4550 at 0.1 mw
Observed
Freq.(MHz)
26563
26569
26577
26604
26620
26633
26634
26647
26666
26674
26777
Literature
Freq.(MHz)
Intensity
(divisions)
1860
2018
1744
74
1669
320
1585
1479
74
60
381
Intensity,%
92
100
86
3.7
82
9.0
79
73
3.7
3.0
19
-------
-18-
Power
1 mw
Table V
CHaSH
OCS 24326 line
= 4550 at 0.1 mw
Observed
Freq.(MHz)
18041
18258
18432
18495
18779
18805
18824
18896
18993
19027
19115
19173
19212
19274
19334
19488
19511
19568
19703
19913
19973
20000
20051
20139
20242
20385
20645
Literature
Freq.(MHz)
Intensity
(divisions)
15
10
40
69
14.6
93
51
24
20
30
5
32
5
12
40
15
49
199
51
55
7
63
55
57
74
66
46
2
1
5
8
2
11
6
3
2
4
1
4
1
1
5
2
6
24
6
7
1
8
7
7
9
8
6
Intensity,%
-------
-19-
Table V
CH3SH (Continued)
Observed Literature Intensity
Freq.(MHz) Freq.(MHz) (divisions) Intensity,%
20929
21056
211O7
21113
2113O
21231
21519
21557
21735
21767
21865
21877
21902
21942
21974
22270
22332
22560
22663
22827
23074
23078
23230
23255
23339
23495
23522
23525
52
65
7
13
55
81
74
66
70
36
149
85
72
72
117
76
97
95
75
225
56
62
148
294
47
57
65
56
6
8
1
2
7
10
9
8
8
4
18
10
9
9
14
9
12
12
9
27
7
7
18
35
6
7
8
7
-------
-20-
Table V
CH^SH (Continued)
Observed Literature Intensity
Freq.(MHz) Freq.(MHz) (divisions) Intensity,'
23565
23618
23774
23805
23989
24068
24453
24485
24879
24989
25009
25O99
25122
25127
25133
25142
25144
25155
25207
25219 r
25999 ซ='25291.87*
25292 ^125290.89*
25562
25661
25918
25999
26076
26144
26348
26388
Frequencies measured at Dow to an accuracy of better than
ฑ0.02 MHz.
151
157
47
95
56
82
26
44
20
19
34
16
29
19
22
44
87
79
225
216
840
36
276
27
25
223
29
34
249
18
19
6
11
7
10
3
5
2
2
4
2
4
2
3
5
10
9
27
26
100
4
33
3
3
27
4
4
30
-------
-21-
Table VI
Acetone
Power = 1.0 mw
OCS 24326 Line = 4550 at 0.1 mw
Observed
Freq.(MHz)
21169
21610
21953
22586
22945
23657
23910
23962
23968
24133
24901
25522
25677
25907
26402
Literature
Freq.(MHz)
Intensity
(divisions)
183
209
209
113
107
245
107
127
194
104
118
132
113
192
127
Intensity,%
75
85
85
46
44
100
44
52
79
42
48
54
46
78
52
-------
-22-
The last column of each table gives the relative intensity of
lines as a percentage of the most intense line in the spectrum
of that compound. We have found such a listing very useful
for qualitative identification work.
Table VII lists the frequencies of the lines selected for the
study. In all cases these were the strongest lines in the
spectra. For CH3SH three lines were selected for study, each
of which represented a different problem as far as adjacent
line interference is concerned. The line at 23230.3 MHz has
a first order Stark effect and no interference from Stark or
other absorption lines near it. The 23256.9 MHz line is the
second most intense line of the spectrum and has several Stark
lines adjacent to it on the high frequency side. The pair of
lines at 25290.8 and 25291.8 were chosen to see how measurements
on such an overlapping pair of lines could be handled. The
Stark components from these lines are fairly well separated
and appear on the high frequency side.
Included in Table VII is a listing of interfering compounds
and the approximate frequencies (ฑ1 MHz) of the interfering
lines. This listing is from data available in our files and
quite probably is not complete. In addition, the possibility
of interference from Stark components of lines at some distance
in frequency from the selected lines can be a serious problem
in quantative work. In many cases qualitative identification
can be made in spite of rather severe interference, while a
quantative analysis is impossible. An examination of a large
catalog of microwave spectra shows that it is essentially im-
possible to find lines with no possible interferences. In a
practical problem, however, many compounds can be eliminated
from consideration from a knowledge of the origin of the problem.
-------
Table VII
Compound
OCS
SO,
CH20
NO,
CH3SH
Potential
Qfl 1 /-ป/"> +~ (~\(\
OL-iCL- LCU
Freq. (MHz)
24325.9
24039. 5
22965.71
266 2O
25290. 8
25291 . 8
Interferences
Freq. (MHz)
24324
24326
24326
24327
24327
24O4O
24O40
24 MO
24041
24 ail
24041
22964
22965
22965
22965
22966
22966
22967
26218
26219
26221
26222
25289
25289
25292
with Selected Lines
Interference
Compound
Die thylamine
Chlorobenzene
Propargyl alcohol
Chlorine tri fluoride
Vinyl bromide
trans- 2 , 3-Epoxybu tane
1 , 3-Dichloropropyne
Thiourea
2-Hydroxy-2-ni tro propion
i trile
Chloromethyl methyl sulfide
Vinyl bromide
4- Methyl pyridine
l-Bromo-2-bu tyne
Ni troe thane
Th iourea
Styrene oxide
Ethanol
Clilorometliyl methyl sulfi
2-Hydroxy- 2-n i tro propion
1 , l-Dichloro-2, 2-di fluoro
Cyclopentanol
a-Bromoothyl benzene
1 , l-Dichloro-2, 2-di fluoro
Ethylene dichloride
2-Hydroxy- 2-ni tro propi on
de
i t r i 1 e
propane
propane
i t. r i 1 e
I
to
to
I
-------
Table VII
Potential Interferences with Selected Lines (Continued)
Compound
CH3SH
CH3SH
NH-
QO 1 c*t~* f*or?
OUxtJC, LUtl
Freq. (MHz)
23256.9
23230.3
23870. 1
Freq. (MHz)
23257
23258
23258
23259
23228
23228
23229
2323O
23230
23231
23231
23231
23231
23231
23870
23870
23871
23871
23871
23872
23872
Interference
Compound
Cyclopentanone
Bromocyclopentnne
Tetrahydrofuran
para Fluorobenzene
Die thylamine
Benzoni trile
l-Bromo-2-bu tyne
para Cresol
cis Dichloroethylene
para Fluorotoluene
1-Bromo bicyclo(2 , 2 , 2) oc tane
2-Hydroxy-2-ni tro propioni tr ile
Chloromethyl methyl sull'ide
Ethyl bromide
Pyrrol idine
2- Pyrrol idone
Acetic acid
1,1, 2-Trichloroc thane
beta Chloroethyl benzene
1-Bromo- 2-bu tyne
Methylene chloride
I
to
-------
-25-
Because of pressure broadening and power saturation effects on
microwave lines, the sensitivity limits of a microwave spec-
trometer must be given for a particular system. Sensitivity
for a compound of low partial pressure in a mixture with air
will be quite different from the sensitivity for the same pres-
sure of the compound in the pure state. Also, the best choices
of microwave power to use for maximum sensitivity in these two
situations may be quite different. For tiie case of the pure
compound at very low pressure, power saturation will limit the
amount of power that can be used. The power limitation in the
compound and air mixture will in general be about one milliwatt.
At approximately this power level, the crystal noise is increasing
as fast as the signal increases and no improvement in signal-to-
noise can be gained by increasing the power level.
Table VIII gives minimum detection limits for each of the selected
compounds as determined from mixtures of the compound with ni-
trogen. The compounds were present in the mixture at the l.O
mole percent level or lower for these determinations. The micro-
wave power was 1.0 mw. The total sample pressure was 100 mTorr.
A scanning rate of one megaHertz per second and a lock-in am-
plifier time constant of 0.3 seconds were used. The N702 and NH3
data were obtained from a flow-through gas system to minimize
adsorption effects. As a result the sensitivity figures for
these compounds may be better than one would obtain in a batch
sample system. Several samples of the CH3SH mixture were in-
jected into the spectrometer before a run was made, in order
to reduce adsorption effects. We consider that the sensitivity
figures given here are useful only as guide-lines.
-------
-26-
Table VIII
Detection Limits in Mole Percent
S02 0.009
CH20 0.008
CH3SH 0.04
Acetone 0. 16
Ammonia 0.002
N02 0.08
-------
-27-
V. QUANTITATIVE ANALYSIS BY THE OPTIMUM
POWER SATURATION METHOD
A. THEORY OF THE OPTIMUM POWER SATURATION METHOD
The quantitative relationships in microwave spectroscopy appear
at first examination to be more direct than for other spectro-
scopic techniques. The absorption of microwave power depends
linearly on the absorption coefficient to a very close approxi-
mation, i.e., AI = 611, where AI is the absorbed power, 6 is
the line absorption coefficient in cm -1 , 1 is the cell length
in cm, and I is the incident microwave power. Two facts com-
plicate the problem: (1) the very small difference in the
molecular populations of the two energy levels involved in an
observed transition permits some degree of power saturation
to occur at a relatively low power level, and (2) molecular
collision broadening of absorption lines (pressure broadening)
is important at typical operating pressures.
Harrington7 has described a procedure which eliminates the prob-
lems of power saturation and pressure broadening. He derived
expressions which show that with a selected amount of power
saturation the intensity of microwave absorption lines depend
only on the partial pressure of the absorbing gas. The effects
of changes in line widths on intensity measurements are eliminated.
Crable and Wahr8 showed that the optimum power saturation pro-
cedure depends only on the detection system of the microwave
spectrometer. The crystal detector output must be proportional
to the square root of the microwave intensity (or power). The
desired detector response can be obtained with special bridge
power balancing equipment available from the Hewlett-Packard
Company. We have demonstrated experimentally that a standard
-------
-28-
broad-banded crystal detector will also provide approximately
the desired response at the relatively high powers required for
optimum power saturation. The response of a crystal detector
to microwave power is essentially linear at microwatt levels
and approaches approximately a square root dependence at milli-
watt levels.
In the application of the optimum power saturation method, the
spectrometer frequency is set to the center frequency of the
desired line. The microwave power level is gradually increased
until the line intensity reaches a maximum value. With increased
power the intensity will decrease slowly; the optimum power
setting is not critical. The observed line intensity at optimum
power is a measure of the number of molecules of that compound
per cm3, or the partial pressure. The actual determination of
partial pressure requires an experimentally determined calibration
curve.
From Crable and Wahr8 the power required at the optimum power
saturation level is
Popt. ~ ~ (Equation 1)
where Av = half line width at half power
_ 16 TT2 < /u../2
B ~ 3 C h2 /UlJ7
C = velocity of light
h = Planck's constant
and /Uij/2 = dipole moment matrix element.
For a complete derivation of B, see reference 9. Equation 1
and the definition of B show that the power required depends
directly on the square of Av and inversely on /Uij/2. The
-------
-29-
dipole moment matrix element contains terms in the rotational
quantum numbers of the levels involved in the transition and
the square of the dipole moment of the molecule.
The absorption coefficient is also related to /u . ./2 since:
An
6Q = -7 = line center absorption coefficient (Equation 2)
Equation 2 is correct for a measurement with no power saturation
and is approximated when low microwave power is used. n is the
number of absorbing molecules per cm3 while A is defined9 as:
A = /U/2 (Equation 3)
where f = fraction of molecules in lower state of the transition
v = center frequency of absorbing line
T = temperature, ฐK
Since the line intensity observed under low power conditions
depends on /\\ . ./2 through A, an order of magnitude estimate of
microwave power required for the optimum power saturation method
can be made from observed line intensities. The most intense
lines require the smallest power to produce optimum saturation.
Also, narrow lines require less power to saturate than broad
lines.
The independence of the optimum power saturation method on line
width is based on the approximation that line width contributions
come only through molecular collisions and wall collisions. For
reasonable pressures this approximation is good. For low total
sample pressure it is no longer valid. The complete form for
the signal, S , observed at optimum power saturation8 is:
max
-------
-30-
Avw+Av .
S = constant n ^ ^ (Equation 4)
max AvD+Avg+Avw+AvM ^
where Av_ = Doppler line broadening contribution to line width
Av,., = Stark modulation contribution to line width
o
Av,y = Wall collision contribution to line width
Av,, = Molecular collision contribution to line width
M
At high pressures, both the numerator and denominator of Equation
4 are dominated by Av,,. At low pressures Av approaches the
value of the other line width contributions, and S no longer
m 3.x
depends only on n, the partial pressure of the desired compound.
B. EXPERIMENTAL RESULTS FOR THE OPTIMUM POWER SATURATION METHOD
The OCS line at 24325.92 MHz was selected for a test of the
optimum power saturation procedure because its high intensity
and narrow line width make it relatively easy to power saturate.
Our line width measurements gave a value of 6.25 kHz per mTorr
of pressure for the molecular collision line width parameter.
Figure 1 shows a plot of OCS line intensity versus OCS pressure.
The data form a reasonable straight line. Note that the low pressure
data deviates from a straight line as discussed above.
The data of Figure 1 did not fit all of our expectations. The
line width, Av, in Equation 1, is proportional to pressure at
high pressures. Thus, according to Equation 1, the power re-
quired for optimum saturation should be proportional to the
square of the pressure. 3.6O mw were required at a pressure
of 20 mTorr and 6.26 mw at 33.5 mTorr. However, the square
of the ratio of these pressures is 2.81, and that times 3.60
mw is 10.1 mw, which is considerably larger than the 6.26 mw
observed. Two factors may explain this result: (1) the operating
-------
-31-
t__, - ~T'~
:::h^fr!7L.::H7T-J
:. ':- y., . . - -7 i: .-.--
- - - ...... - - 1 ..... T '
, | - : | ..... - I -----
:..,.. :.
.;_!..:
r ; ( --.,--
&4:4.-
.^_4^:J^.L::^._Lu::,:-
iiiSH^E'l^W-- ^::-
r:^:fei;;d-.[:::;[.::;i;--
-I ---
:T.: ;
--
Trrj-.-r;:
f--
.f . ::
ii:Tn
"F7"
;| r
- :
_^_ -^_ ^- _ 1^
ILL_L:
Sfflr
-~t ): * " -rt-:-'
: . -~l .T7T| I"7 ~~T 1 LI !7
;uiie_^L
i.
fL^&^l:^!^^ii::rl-:J
fTH:.!;-.
.
i:--: ft-
. . ._ .
... i,
rF~r
-------
-32-
characteristics of our detector do change with power and (2)
the very narrow molecular collision line width parameter makes
the required approximation that the molecular collisions dominate
the line width be somewhat poorer than desired. Using a Stark
modulation frequency of 10O kHz the observed half line width
at half power was 205 kHz at 20 mTorr of which molecular col-
lisions contributed only 125 kHz.
Better results are obtained from Equation 1 if actual measured
line widths are used. At a Stark frequency of 100 kHz our
measured half line widths were 205 and 278 kHz at pressures
of 20 and 33.5 mTorr, respectively. The square of the line
width ratio is then 1.84 and predicts, from the 20 mTorr data,
a required power of 6.6 mw at 33.5 mTorr. This is in reasonable
agreement with the 6.26 mw observed.
As part of our evaluation it was necessary to have a means of
estimating the microwave power required for the optimum power
saturation method for a range of molecules. For this reason the
following expression was derived. From Equation 2 an expression
for the observed line intensity can be given as:
Center line intensity = C = -r (Equation 5)
A CAv
or A =
The microwave power term has been left out of Equation 5. It
is assumed here that the line intensities are measured at a
power level of 0.1 mw to minimize power saturation effects.
From the definitions of A and B
D A 2kT CAv 2kT ,_ ,. >.
B = A = . (Equation 6)
h2fv 2 n h2fv 2
o o
-------
-33-
Substituting Equation 6 in Equation 1, we obtain
? =/CAv "^ \ = " ฐ ' constant (Equation 7)
V~~n~
If Av is substituted for n(since Av is proportional to n) and,
further, f is lumped into the constant, Equation 7 becomes:
P = constant x vuv^ vo (Equation 8)
From the OCS data of 3.60 mw and a Av of 205 kHz, the constant
is 1.12 x 1O 9 and Equation 8 becomes:
-, 10 v ,n-9 [Av(kHz)]2 [vn(MHz)]2 OCS Ints.
P(mw) 1-12 X 1U C ' 4550
(Equation 9)
The OCS Ints. is the observed line intensity in divisions of
the OCS 24325 MHz line for a microwave power of 0.1 mw and a
pressure high enough to be in the pressure broadened region.
The last term of the equation is included for approximate stan-
dardization of different spectrometers. The Av here is the
observed half line width at half power measured or calculated
at the sample pressure of interest in the analysis. If the
line width parameter L is known, then the line width can be
calculated by multiplying L by the pressure in units compatible
with L. Our experimental work showed that only for OCS with
its very narrow collision broadened line width was the line
width broadening from Stark modulation of significance.
In gas mixtures, the actual line width of the desired line in
the mixture must be used. The OCS measurement discussed above
required 3.60 mw for 20 mTorr of pure OCS. A partial pressure
of 20 mTorr of OCS in a mixture at a total pressure of 100 mTorr
-------
-34-
would require considerable more power to produce optimum power
saturation. The amount of power would depend on the nature of
the other components in the mixture with respect to their col-
lision rates with OCS molecules.
Table IX gives the microwave power needed for optimum saturation
of a number of typical molecules. These data were calculated
from Equation 9 for a sample of the pure compound at a pressure
of 20 mTorr. Although these calculated results should only be
considered as good estimates, they do show that relatively high
microwave powers are required even for molecules which are
usually considered to be strong microwave absorbers. Since
the microwave power available in a K-band spectrometer is of
the order of 1O raw, and is lower yet for spectrometers oper-
ating at higher frequencies, the large power requirement places
a severe limitation on the general application of the method.
In the only published work of an application of this technique,
Funkhouser10 et.al.found that insufficient power was available
in R-band to make measurements above 20 mTorr total pressure
of a 10.1 volume percent sample of acetone in nitrogen.
The application of the optimum power saturation method to air
pollution samples is somewhat more favorable. Such samples
will consist of small concentrations of gases of interest in
the presence of large concentrations of air. The effect on
line width of collisions between the microwave absorbing mole-
cules and nitrogen or oxygen molecules is small by comparison
with the effect of collisions among the absorbing molecules
themselves. Thus, the microwave power required is reasonable -
of the order of a few milliwatts. Although the optimum power
saturation technique is applicable to air pollution samples,
we believe that simpler and easier to use techniques are more
practical, particularly for this special case. These techniques
are discussed in later sections.
-------
-35-
Table IX
Microwave Power Required for
Optimum Power Saturation
Line Width Power Required
Parameter Line Intensity at 20 mTorr
(kHz/mTorr) (div) (mw)
CH20 22965. 57 MHz
24068.35
S02
CHF3
CH3CF3
CH3COCH3
20335.
24083.
2O697.
20740.
21169.
23656.
43
46
73
53
95
28
18.0
17.3
33.3
49.8
52.1
50.4
24.7
20.1
18.0
17.3
33.3
49.8
52.1
50.4
982
831
721
550
275
183
245
150
130
83
140
90
300
260
C. SUMMARY AND CONCLUSIONS ON THE OPTIMUM POWER SATURATION METHOD
The optimum power saturation technique allows line intensities
to be used as a linear quantitative measure of concentration,
or partial pressure. The line intensities at optimum power
saturation are independent (at sufficiently high total gas
pressures) of all line broadening effects. A calibration curve
is required for each compound to be determined.
Disadvantages are: (1) high microwave powers are required and
(2) each measurement requires a separate optimization of the
microwave power. The high microwave power requirement eliminates
the application of this technique to many molecules, if not most.
For general applications of microwave spectroscopy to quantita-
tive determinations, other techniques, which follow, appear
more attractive.
-------
-36-
VI. QUANTITATIVE ANALYSIS BY LINE AREA MEASUREMENTS
AND LINE HEIGHT MEASUREMENTS
A. DISCUSSION OF QUANTITATIVE ANALYSIS BY AREA MEASUREMENTS
A second method of quantitative analysis uses microwave ab-
sorption line areas as a measure of partial pressures of a
compound. Townes and Schawlow9 derived the following ex-
pression for microwave absorption coefficients.
6 - 1 "V*' ' (v-v0/: ปv)ป (Equation 10)
where v = microwave frequency and VQ = line center frequency.
All other terms were defined earlier. They9 showed that after
integration over frequency, Equation 10 became:
"ซ'' -V
(Equation 11)
= constant x n
The constant of Equation 11 applies to a specific molecular
absorption line and temperature. Since the observed line
intensity is microwave power, I, times 6, the integrated ob-
served line area is:
Integrated line area = constant x I x n (Equation 12)
Thus for a fixed power level the absorption line area is linearly
proportional to the partial pressure. It is assumed that the
power level is low enough to minimize power saturation effects.
For molecules other than very strong absorbers this condition
is easily satisfied.
-------
-37-
An examination of Equation 10 suggests a simpler method of
analysis without requiring the direct measurement of line
areas. When Equation 10 is written for the line center fre-
quency, it becomes:
5 = ^/2 ' (Equation 13)
If Equation 13 is multiplied by the line width, Av , it then
becomes independent of line width, or:
Center line intensity x Av = constant x I x n
(Equation 14)
Equations 12 and 14 show that partial pressures can be determined
from either a measurement of line area or the product of the
line intensity and the half line width at half power. The line
area measurements are somewhat cumbersome since it is not prac-
tical to integrate the entire frequency range experimentally.
Line width and intensity measurements can be measured with
reasonable ease if the frequency scanning speed and chart
paper speeds of the spectrometer are selected to give a line
width of 10 to 15 cm on the chart.
B. QUANTITATIVE ANALYSIS FROM LINE HEIGHT MEASUREMENTS
For certain special cases, quantitative results based on a
simple intensity measurement are valid. Such cases occur when
a minor component is to be determined in a mixture in which
the major part of the mixture remains relatively unchanged in
concentration and composition. Air pollutants will in general
satisfy this requirement. For example, the line width of a
small percentage of S02 in air is determined primarily by col-
lisions between S02 molecules and oxygen and nitrogen molecules
-------
-38-
in the air. Over the concentration range of 0 to about 5 mole
percent S02, the S02 line width will be essentially a constant.
Thus, the line intensity becomes a direct measure of mole per-
cent S02 at low concentrations in air.
Another interesting and practical characteristic of this special
case is the fact that for a given minor component concentration,
the observed line intensity is independent of the total pres-
sure. An increase in total pressure produces an increase in
number of absorbing molecules per cm3 (the minor component)
which causes a proportionate increase in line area. In addition,
the proportionate increase in the number of major component
molecules results in an increased collision rate with the ab-
sorbing molecules and a proportionate increase in line width.
The net result is a line with increased area and line width,
but a line with an unchanged intensity.
C. EXPERIMENTAL RESULTS
Figure 2 is a plot of line height (Intensity) and area measure-
ments for the OCS 24325.9 MHz line. The line intensity plot
shows good linearity to 10 mTorr and reasonable linearity to
approximately 20 mTorr. The straight line drawn through the
area points shows that a good linear dependence exists between
line area and partial pressure of OCS.
The line height and product of line height and width for the
S02 24039.6 MHz line are shown in Figure 3. The line height
curve shows reasonable linearity at pressures below 10 mTorr.
The product of height and width show good linearity.
Data for CH20 are given in Figure 4. The line height is again
reasonably linear below a pressure of 10 mTorr. The plot of the
-------
-39-
-------
-40-
:.::;.. I -.., i .
_.^_L^;.j.;_--.U-_: ii:: i ..r: >..!_.. i _.- i;-^:
>/t;..-: Operating Conditions
t> ..;.
- i r :rr:"/^~^'--~^ Spectrometer
Power
constant
- : ::-G;,
r..\]-:[/X:X:X:JX Stark volt.
TTฃ
0.1 mw
0.1 sec
7.5 MHz/min
1OOO v at
100 kHz
:t:
-[---.-r-
iEi'guire :B
i - r
- x x ; - x j x '
-------
-41-
-------
-42-
product of height and width is a good straight line from about
10 to 90 mTorr. The low pressure end shows a slight curvature
and did not appear to extrapolate to zero for zero pressure.
The high pressure end (additional higher pressure points were
not plotted here) shows a downward curvature toward the pres-
sure axis. This probably results from overlap of the absorption
line and Stark lines. At high pressures the Stark components
for most of the compounds studied were broad enough to overlap
the absorption line with a resulting decrease in line intensity.
Figure 5 gives plots of line heights and height x width products
versus pressure for the acetone 21169.2 MHz line. The line
height is linear at pressures below 15 mTorr. The product of
height x width shows considerable scatter even though the points
plotted are the average of several determinations. Some of the
variation results from measurement problems because of lowered
signal-to-noise for acetone. Another source of variation is
the possibility of Stark component overlap from nearby lines.
Line height data for the three CH3SH lines at 25291, 23256.9,
and 23230.3 MHz are shown in Figure 6. Line heights for the
doublet at the nominal frequency of 25291 were measured as the
intensity of either line at low pressures where they are com-
pletely resolved. At higher pressures the highest peak, or
the combined peak, was measured. The plot for the 23256.9 MHz
line is quite normal with a linear portion below 10 mTorr. The
plot of the 23230.3 MHz line does not appear to have a really
linear portion at low pressures. The plot of the 25291 doublet
has a linear portion below approximately 10 mTorr. Above 10
mTorr the general shape of the curve is quite different from
the usual line height plot. The explanation is the combining
of the two doublet components to form a single central peak at
higher pressures.
-------
-43-
-------
----- . - . r -
','.', : . : j ' " : '. ' l ;
-44-
^-GHi&tyrT--'.'.
\i i -i -i
~
::.
I/'!. ,: ! . Operating Conditions
[__Jl_; Spectrometer #2
T. f: Power % 1.0 mw
; . Power (25291) 0.4 mw
J_jj__!.- RC constant 0.03 sec
!! [::;.;|;;: Scan rate 0.7 MHz/sec
]- Stark volt. lOOOv at
:t..-.J.--L. 100 kHz
'_-:!r::;l'.--!-1 -'I:":i:-:!::';r7:- "7- 77 r; -| ::. -:'.
i_. J_.
'
C.
f : ih:;?!!
:.t:-~
-7.-
:j^:i:6ia;j^^:iigb:^i7i^^ini
5-ufeVf'n'! :/M/777:7^rr:r;-^:-!:-^ ::-|;-::
~ .. T^ - .~j..__.. . 1 ,_! ... 1 ', j r I ._ J ' '
L J^|:_.. | :: 1 '
-------
-45-
Figure 7 shows the dependence of the products of height and
width on pressure for the three CH3SH lines. The plots for
the 23256.9 and 23230.3 MHz lines are linear with pressure.
The plot for the 25291 MHz line has a good linear portion over
a pressure range of 1O to 80 mTorr. The line width used for
the 25291 line was the entire width of the doublet, i.e., from the
lower side of the low frequency component to the higher side
of the high frequency component. The inflections in this plot
are certainly a result of the method of measuring line width.
The overall result obtained from these data is that area or
the product of line height and width is a linear function of
pressure. The line height is linearly related to pressure at
low pressures. The data for the 25291 MHz doublet of CH3SH
also approximates these relationships.
-------
-46-
i-----!~i -.--Heici hr, X : i///dth - X '--Pr&ss-L
1 ' \ -.-, cj ; i , . ..; : ! !.'.; i
_J Operating Conditions
Spectrometer
Power
Power (25291)
RC constant
Scan rate
Stark volt.
1.0 mw
0.4 mw
0.03 sec
O.7 MHz/sec
lOOOv at
100 kHz
-------
-47-
VII. GAS BLENDS AND THE ADSORPTION PROBLEM
A gas blending system was constructed of glass and greaseless
stopcocks. Gas volumes were measured In a calibrated buret at
atmospheric pressure. Measured gas volumes were passed into
a large mixing chamber by mercury leveling bulbs. After the
gases were in the mixing chamber, a metal mixer in the chamber
was raised by an external magnet and allowed to drop through
the gas mixture. A spring at the bottom of the mixer cushioned
the fall of the mixer at the bottom of the chamber.
The blending system and the blending technique were checked by
preparing three blends of argon in nitrogen and then determining
the actual concentrations with a CEC 21-104 mass spectrometer.
Argon and nitrogen were used in this test to eliminate all ad-
sorption problems. The results obtained are given in Table X
below.
Table X
Argon in Nitrogen Blends
Blended Mass Spectrometer
Composition Analysis
mole % mole % Error,%
O.102 0.105 3
1.02 1.02 0
10.01 10.14 0.3
These results show that accurate blends can be prepared with
this gas blending system.
Three blends of S02 in nitrogen were carefully prepared on the
gas blending system. The blended compositions along with mass
-------
-48-
spectrometer analyses of these mixtures are given in Table XI.
Mass spectrometer analysis no. 1 was performed within two hours
of the blend preparation, while mass spectrometer analysis no. 2
was run on the following day. Although the blended samples
were stored in glass Shepard traps with greaseless valves, the
mass spectrometer results indicate that S02 was being adsorbed
on the walls of the container as a function of time. Included
in Table XI are line heights observed for the S02 25392 MHz
line from the three S02 blends. A total sample pressure of
1OO mTorr and a power of 0.1 mw were used. The ratios of these
line heights are in relatively good agreement with the blended
composition. If the 10.0 mole % blend is taken as a standard,
then percentages calculated for the other two blends based on
a linear dependence of height on partial pressure are 0.97 and
0.083. A comparison of the mass spectrometer and microwave
spectrometer results suggest that some S02 was lost from the
samples through adsorption on container walls, and that possibly
adsorption in the inlet of the mass spectrometer is more severe
than adsorption in the microwave spectrometer. The overall
result is a demonstration of the problems of preparing accurate
gas blends of gases which adsorb. S02 is one of the easier
compounds to handle in a microwave spectrometer with respect
to adsorption.
Table XI
Blends of S02 in Nitrogen
Blended Mass Spectrometer Mass Spectrometer
Composition Analysis #1 Analysis #2 S02 25329 MHz
mole % mole % mole % Height (in.)
0.100 0.052 0.023 0.47
1.00 0.76 0.68 5.48
10.01 9.82 9.33 56.4
-------
-49-
Table XII gives information on adsorption problems as related
to repeated sample introductions. Four successive introductions
of the 0.1 mole % blend of S02 into the spectrometer caused
the line height to increase from 0.37 in. to 0.50 in. A 10.0
mole % S02 blend was then introduced, scanned, and pumped out.
The next introduction of the 0.1 mole % S02 blend gave a line
height of 1.7 in., or 3.4 times as high as the last run before
introducing the 10.0 mole % blend. The high value, of course,
results from desorbing of S02 which had been adsorbed from the
10.0 mole % blend. These data show that the results obtained
on a microwave spectrometer are highly dependent on the previous
history of the spectrometer cell. This problem could be over-
come by having a number of absorption cells available and ar-
ranged so that the unused cells were pumped constantly to remove
adsorbed gases.
Table XII
S02 in Nitrogen
Total Pressure = 100 mTorr
Power = 0.1 mw
Sample
mole %
O.10
0.10
0.10
0.10
10.O
0.10
Run No.
1
2
3
4
5
6
Time,min
O
9
18
32
45
58
Line Intensity,
in.
0.37
0.45
0.48
0. 50
60.7
1.7
Similar experiments were carried out with CH3SH and with similar
results. The cell became conditioned for CH3SH for some days
after it had been exposed to CH3SH for an hour. After a high
-------
-50-
pressure exposure to CH3SH, small concentrations of CH3SH could
not be determined with accuracy. Nitrogen was introduced into
the spectrometer and the cell shut off from the inlet system.
After five minutes a scan for CH3SH showed a small line from
CH3SH desorbed from the walls.
Figures 8 and 9 show the decrease in CH3SH line height as a
function of time from experiments with spectrometers no. 1 and
no. 2. Each spectrometer was relatively clean of CH3SH since
the compound had not been in the spectrometers for several weeks
prior to these runs. On a percentage basis, the adsorption in
the brass cell of spectrometer no. 1 was worse than the stain-
less steel cell of spectrometer no. 2. However, these limited
data should not be interpreted as a claim for a stainless steel
cell as a means of reducing adsorption. Earlier experiments of
ours with methyl alcohol indicated that the stainless steel cell
adsorbed more rapidly than the brass cell. The CH3SH data for
spectrometer no. 2 was obtained at a higher pressure than that
of spectrometer no. 1. Thus, the number of CH3SH molecules ad-
sorbed in the stainless steel cell was higher. Once the cell
walls become conditioned, the rate of adsorption is reduced.
The work reported above in this section illustrate a major prob-
lem in the determination of concentrations of gases in small
concentrations in a mixture. To handle such samples a technique
for the elimination of the effects of adsorption must be devised.
One such attempt, the use of a flow-through cell, is reported
in the next section.
-------
-51-
-------
-52-
-------
-53-
VIII. MEASUREMENTS WITH A FLOW-THROUGH CELL
A. DESCRIPTION OF THE EQUIPMENT AND PROCEDURE
Possibly the most difficult problem to overcome in the analysis
of low concentrations of a gas in a mixture is the adsorption
of the gas in the walls of the cell. A Stark cell of reasonable
sensitivity usually has a cell 10 feet in length. The surface
of the cell walls and the surface of the metal septum down the
center of the cell constitute a large area for adsorption. In
an attempt to overcome this problem the following experiments
were carried out on a flow-through gas cell.
The stainless steel cell of spectrometer no. 2 was originally
designed with pump-out ports at both ends of the cell. By
attaching a vacuum system to the end of the cell away from
the inlet system, a flow-through system was obtained. The
vacuum system consisted of the combination of a cold trap, oil
diffusion pump, and a fore pump. The vacuum system was con-
nected to the cell through a valve and a leak. The leak which
was finally chosen was a metal plate drilled with a no. 80
drill, i.e., a 0.0135 inch diameter hole. With this leak in
place, the flow out of the cell is 233 mTorr cm3/sec.
The first system for continuous mixing of gases used gas burets
to determine gas flow rates and thus determine the proper set-
tings of a pair of needle valves. This system was used in at-
tempts to mix small amounts of S02 with nitrogen. Difficulties
with the burets, and in maintaining the valves at a fixed posi-
tion for the very low SO2 flows caused us to discard this system.
The second system used rotameters to set the flow rates of the
two gases. With a mixture flow rate of greater than 500 cm3/min,
-------
-54-
the smallest SO2 rate required was several cms/min. This rate
could be monitored with the rotameter, and the Nypro valve used
did not change its setting with time. In this system, the
nitrogen and SO2, or other gas, were each passed through a
valve and rotameter. The outputs from the two rotameters were
joined by a Y connection and the gas mixture brought to the
spectrometer introduction system through approximately 30 feet
of 1/4 inch plastic tubing. A tee connection at the spectrometer
with one leg open to the atmosphere and the other leg connected
to the spectrometer through a valve, provided for sampling at
constant atmospheric pressure. The gases exhausted from the
tee were returned to the hood by means of a blower and a 4 inch
exhaust line.
Pressure in the 5 liter reservoir of the sample introduction
system was maintained at one Torr by properly setting the leak
valve between the reservoir and the cell. A by-pass valve around
the leak valve was opened initially to bring the cell pressure
up to 100 mTorr quickly. Once the cell was pressurized and
the flow-through system adjusted, the pressure in the cell was
quite stable.
B. EXPERIMENTAL RESULTS WITH THE FLOW-THROUGH SYSTEM
Gas concentrations were based on the readings of the calibrated
rotameters. A point of experimental interest was the time
required to equilibrate the system. The time required for a
sample to pass through the plastic connecting tube was found
to be 1O to 12 seconds. Measurements made with S02 and nitro-
gen mixtures are given in Table XIII. Line heights in chart
divisions for the 22482.5 MHz line, and the time required for
the line height to reach a maximum after each change in S02
concentration are given. The times given are good to about one
minute. These data show that for this particular system, about
15 minutes is required to reach equilibrium.
-------
-55-
Table XIII
Height of SO;, 22482.5 MHz Line and
Time to Reach Maximum Amplitude
Line Height
Mole % SOg (Chart Divisions) Time (minutes)
0.037 8.5 not measured
0.087 12.5 -5.5
0.15 17.0 6.5
0.304 28.0 10
0.622 53.5 12
1.O5 89.5 -10
2.45 191 12
4.98 327 9
9.90 555 10
Figure 10 is a plot of the line height data of Table XIII. Again,
these results show that for low gas concentrations an essentially
linear relationship exists between line height and S02 concentra-
tion. The direct line height versus concentration procedure is
not only the easiest technique to use, but it also does not have
non-linearity problems at the low concentration range common to
the other procedures.
Data from the flow-through system for CH3SH in nitrogen are shown
in Figure 11. These results for the 25291 MHz doublet are not
as linear as one would like. The problem here is probably a
result of overlap of the two lines with increasing pressure.
From these experiments, we estimated that the least detectable
concentration was 0.05 mole percent at a power of 0.1 mw in the
stainless steel flow-through system.
-------
-56-
-------
-57-
_ -- .- -- -_ . _ . .. _ ._._ _;...._ __ -..'... _ -- .-;-.- .
JIliliiB^
' ""
-------
-58-
Acetone in nitrogen mixtures were prepared by first passing
nitrogen slowly through liquid acetone to provide a saturated
acetone in nitrogen stream. This mixture was then further
diluted with additional nitrogen. Acetone concentrations were
based on vapor pressure data as a function of the temperature
of the acetone. Data for the 21170 MHz line of acetone at a
power of O.I mw is shown in Figure 12. The low intensity of
the acetone line results in some scatter of these data. An
indication of the approximate noise level is given. A reasonable
straight line plot was obtained. From measurements over a
range of operating conditions, we estimated that a concentration
of 0.05 mole percent of acetone in nitrogen can be detected at
a power level of 3 mw with this system.
The problem of producing gas mixtures of formaldehyde and nitro-
gen was a challenge since formaldehyde does not normally exist
as a gas at higher pressures. The formaldehyde was produced
by gently heating paraformaldehyde in a small furnace through
which nitrogen was passed. The concentration of formaldehyde
was determined from the weight loss of paraformaldehyde per
unit of time and the measured flow rate of nitrogen. Although
we observed no direct evidence that we later lost formaldehyde
through polymerization, our calculated concentrations may be in
error on the high side if polymerization did take place. Figure
13 shows plots of the formaldehyde 22965.7 MHz line at three
different power levels. Reasonable straight lines were obtained,
particularly considering the potential errors in the mixing
process for this system. At a power level of 5 mw, the minimum
detectable concentration was estimated to be 0.002 mole percent.
The high adsorption of NH3 and its high sensitivity caused spe-
cial problems. In order to dilute the NH3 sufficiently to
determine detection limits, a double dilution was required to
-------
-59-
^
' ~ = " ';'' ' : ~ '
*
L:-L--T:.-::
tmm-r
::p J-a^^
-------
-60-
-------
-61-
keep rotameter readings on a usable portion of their scales.
Measurements made of equilibrium times indicate that times of
the order of 6 to 8 hours are required to completely equilibrate
the system. These are only estimates since the long times in-
volved introduced considerable error into our measurements.
Figure 14 shows data obtained at a power level of 5 mw. Times
allowed for equilibration for each point are also given. This
plot is definitely not a straight line because of adsorption
problems. The least detectable concentration was estimated to
be of the order of O.OO2 mole percent at a power level of 5 mw.
One blend of N02 at the O.I mole % level in nitrogen was studied.
This mixture was prepared using the standard rotameter procedure.
The mixture was allowed to flow through the plastic tubing for
two hours to equilibrate the tubing before admitting gas to the
spectrometer. When the mixture was admitted to the spectrometer,
a slow increase in the intensity of the 26569.21 MHz line occurred
with time. From these data, we estimated that approximately 6
hours time would be required for the line to reach 99% of its
maximum intensity. A portion of these data are shown in Figure
15. Our conclusion after a number of attempts to obtain better
data is that a microwave spectrometer is practically useless to
obtain quantitative results for low concentrations of N02. Under
strictly controlled conditions, quantitation may be possible,
but the effort and time required are excessive. The least de-
tectable concentration is something less than 0.1 mole %.
C. CONCLUSIONS
The results reported above show that the increased complexity of
the flow-through system gains very little in the ease and accuracy
of handling compounds which adsorb readily on cell walls. Es-
sentially the same results can be obtained by preconditioning
the cell through several batch introductions of the same sample.
-------
-62-
-------
-63-
-------
-64-
IX. SUMMARY
Background information and theory, particularly as they apply to
analytical applications are presented. Three possible analytical
procedures are discussed in some detail. These are the optimum
power saturation method, line areas, and line intensities.
The optimum power method has the advantage that the problem of
varying line widths is completely eliminated at pressures above
some minimum value. It has two important disadvantages: (1) the
power level must be optimized for each line and sample pressure
encountered, and (2) the amount of power required is well beyond
the capabilities of most microwave sources for most molecules.
Provisions must be made to operate the crystal detector of the
spectrometer so that its output is proportional to the square
root of the applied power.
The line area is proportional to the partial pressure of an ab-
sorbing molecule and is also independent of the line width. It
is however somewhat tedious to measure. We have shown that the
product of the line height and the half power width has the same
properties as the line area. Measurement is relatively quick
and easy. A disadvantage of both area methods is that the scan-
ning rate of the spectrometer must be uniform and repeatable.
This was accomplished by locking the frequency of the source to
a harmonic of a "tuned" crystal oscillator.
Direct quantitation by means of the line intensity is a very
simple and adequate method for certain cases of interest. These
include any system in which the desired component is in low
concentrations and the composition of the bulk of the mixture
is relatively constant. Trace components in air is an ideal
system for this method.
-------
-65-
The problem of adsorption of gases on the walls of the absorption
cell is serious. The amount of such adsorption to be expected
was shown by several experiments in which gas pressure was mea-
sured as a function of time. Equipment was assembled and a
study conducted to determine whether the adsorption problem
could be eliminated by a continuous flow-through system. The
results indicated that strongly adsorbing compounds can be
handled easier through preconditioning of a standard batch cell.
-------
66-
X. CONCLUSIONS
Qualitative and quantitative analyses can be performed on a
microwave spectrometer. The outstanding advantage of microwave
spectroscopy is the uniqueness of the spectra obtained. For
problems which require the positive identification of a compound
which has a good microwave spectrum, a microwave spectrometer
is very useful. For essentially all other analytical problems,
other forms of spectroscopy, such as mass spectrometry and
infrared, are faster, cheaper, and considerably more versatile
as far as the range of compounds that can be examined. The
sensitivity of a microwave spectrometer is not outstanding ex-
cept for a few compounds like ammonia. Our final conclusion
is that for air pollution work a microwave spectrometer should
be considered as a very specialized analytical tool to be used
only in those specialized cases in which its unique identification
ability can be utilized.
-------
-67-
REFERENCES
1. Townes, C. H., A. N. Holden and F. R. Merritt, Phys. Rev.
_74, 1113 (1948).
2. Lov/, W. and C. H. Townes, Phys. Rev. 75, 529 (194S) .
3. Dakin, T. W., W. S. Good and D. K. Coles, Phys. Rev. 71,
640 (1947).
4. Townes, C. H. and S. Geschwind, Phys. Rev. 74, 626 (1948).
5. VanVleck, J. H., Phys. Rev. Tl, 413 (1947).
6. Bragg, J. K., and A. H. Sharbaugh, Phys. Rev. 75, 1774
(1949).
7. Harrington, H. W. , J. Chem. Phys. 4_6, 3698 (1967).
8. Crable, G. F. and J. C. Wahr, _J. Chem. Phys. 51, 5181 (1969)
9. Townes, C. H. and A. L. Schawlow, Microwave Spectroscopy
McGraw-Hill, New York (1955).
10. Funkhouser, J. T., S. Armstrong and H. W. Harrington, Anal.
Chem. 40, 22A (1968).
------- |